Robustness of Magic in Quantum States

• Robustness of Magic is a resource measure that quantifies how far a quantum state deviates from a stabilizer mixture and underpins quantum speedup.
• It evaluates magic persistence under various noise models using efficient witnesses like stabilizer Rényi entropy and log-free robustness.
• Scalable algorithms and experimental validations highlight its significance in classical simulation overhead, fault tolerance, and cryptographic applications.

Magic, in the context of quantum information theory, refers to the nonstabilizerness of quantum states—i.e., the property of a quantum state that is not a convex mixture of stabilizer states, and thus is essential for enabling quantum speedup and universal fault-tolerant quantum computation. The robustness of magic (RoM) is a class of monotonic resource measures quantifying how much a given quantum state or operation deviates from the stabilizer set. Robustness encapsulates the persistence of nonstabilizerness under various physical and computational processes, particularly in the presence of noise, decoherence, and other resource-degrading mechanisms. Recent developments reveal that magic exhibits remarkable resilience, closely linked to both quantum computational power and structural aspects of quantum many-body systems.

1. Formal Definitions and Magic Witnesses

The gold standard quantitative measure is the robustness of magic, defined for an $n$-qubit state $\rho$ as the solution to the convex program: $$R(\rho) = \min \left\{ \sum_i |x_i| : \rho = \sum_i x_i |S_i\rangle\langle S_i|,\, \sum_i x_i = 1 \right\},$$ where the $|S_i\rangle$ are pure stabilizer states spanning a superexponentially large set (Howard et al., 2016, Hamaguchi et al., 2023, Timsina et al., 17 Jul 2025). The value $R(\rho)=1$ iff $\rho$ is a stabilizer mixture; $R(\rho)>1$ signals magic.

Because RoM is computationally intractable for large $n$, efficient witnesses based on the stabilizer Rényi entropy, $\mathcal{M}_\alpha$, are employed. For example, for $\alpha \ge 1/2$,

$\rho$0

where $\rho$1 is the $\rho$2-Rényi entropy. $\rho$3 certifies nonstabilizerness (Haug et al., 25 Apr 2025).

Further, log-free robustness $\rho$4 offers an operationally significant lower bound on simulation cost and is closely tied to the stabilizer fidelity monotone $\rho$5; $\rho$6.

2. Robustness of Magic Under Noise

Magic is generically robust to several classes of noise. Under global depolarizing noise,

$\rho$7

for $\rho$8 Haar random, the filtered witness $\rho$9 remains $$R(\rho) = \min \left\{ \sum_i |x_i| : \rho = \sum_i x_i |S_i\rangle\langle S_i|,\, \sum_i x_i = 1 \right\},$$0 as long as $$R(\rho) = \min \left\{ \sum_i |x_i| : \rho = \sum_i x_i |S_i\rangle\langle S_i|,\, \sum_i x_i = 1 \right\},$$1: even as the depolarizing rate approaches $$R(\rho) = \min \left\{ \sum_i |x_i| : \rho = \sum_i x_i |S_i\rangle\langle S_i|,\, \sum_i x_i = 1 \right\},$$2 exponentially quickly with system size, detectable magic persists for sub-exponential (in $$R(\rho) = \min \left\{ \sum_i |x_i| : \rho = \sum_i x_i |S_i\rangle\langle S_i|,\, \sum_i x_i = 1 \right\},$$3) noise strengths. For local depolarizing noise, circuit depth, rather than system size, determines the critical threshold: in random local circuits, a critical depth $$R(\rho) = \min \left\{ \sum_i |x_i| : \rho = \sum_i x_i |S_i\rangle\langle S_i|,\, \sum_i x_i = 1 \right\},$$4 exists such that observable magic survives up to $$R(\rho) = \min \left\{ \sum_i |x_i| : \rho = \sum_i x_i |S_i\rangle\langle S_i|,\, \sum_i x_i = 1 \right\},$$5, independent of $$R(\rho) = \min \left\{ \sum_i |x_i| : \rho = \sum_i x_i |S_i\rangle\langle S_i|,\, \sum_i x_i = 1 \right\},$$6 (Haug et al., 25 Apr 2025).

In many-body systems, hypergraph states with high-degree interactions feature vanishing magic thresholds as $$R(\rho) = \min \left\{ \sum_i |x_i| : \rho = \sum_i x_i |S_i\rangle\langle S_i|,\, \sum_i x_i = 1 \right\},$$7, but families with "global" magic support can have non-vanishing thresholds independent of $$R(\rho) = \min \left\{ \sum_i |x_i| : \rho = \sum_i x_i |S_i\rangle\langle S_i|,\, \sum_i x_i = 1 \right\},$$8, indicating noise-resilient macroscopic magic (Wei et al., 2024). This decouples robust global magic from locality: even if all $$R(\rho) = \min \left\{ \sum_i |x_i| : \rho = \sum_i x_i |S_i\rangle\langle S_i|,\, \sum_i x_i = 1 \right\},$$9-qubit marginals are nearly stabilizer, the global state can retain substantial magic.

3. Computational and Experimental Methodologies

Computing or detecting magic for large systems necessitates scalable algorithms and reliable proxies. For moderate $|S_i\rangle$0, column-generation and fast Walsh–Hadamard transform subroutines compute RoM exactly or approximately up to $|S_i\rangle$1, leveraging symmetries, overlap-based screening, and permutation reductions to avoid storing or enumerating the superexponential stabilizer set (Hamaguchi et al., 2023). For larger subsystems or matrix product states, stabilizer Rényi entropy witnesses and property-testing protocols provide efficiently computable, sample-optimal means for magic detection and quantification (Haug et al., 25 Apr 2025).

Experiments, such as those on IonQ hardware, employ two-copy Bell measurements to access $|S_i\rangle$2, demonstrating persistent mixed-state magic even when the global depolarizing fidelity is as low as $|S_i\rangle$3. These methods allow robust certification in realistic, noisy, and shallow-circuit regimes (Haug et al., 25 Apr 2025).

4. Phase Transitions and Many-Body Scaling

Random stabilizer codes with coherent errors display a phase transition in their magic content as a function of error strength: below a critical threshold, syndrome measurements eliminate circuit-accumulated magic, while above it magic accumulates in a "concentrated" manner (Niroula et al., 2023, Wei et al., 2024). This parallels the behavior of entanglement and order parameters in quantum phases but is governed by constraints of the stabilizer polytope rather than solely by local observables.

In quantum critical systems such as the transverse-field Ising chain, the mutual robustness of magic between partitions exhibits power-law decay at criticality and an algebraic scaling of critical temperature with subsystem size. Notably, magic correlations persist at finite temperature and do not undergo "sudden death" like entanglement measures—magic provides a more robust diagnostic of quantum criticality (Timsina et al., 17 Jul 2025).

5. Operational Significance and Cryptographic Aspects

Robustness of magic directly quantifies the overhead for classical simulation of near-Clifford circuits: the sample complexity of Monte Carlo schemes scales quadratically with $|S_i\rangle$4 (Howard et al., 2016, Seddon et al., 2019). Efficient property-testing of $|S_i\rangle$5 distinguishes low-magic versus high-magic states with poly$|S_i\rangle$6 samples provided $|S_i\rangle$7. States with high entropy $|S_i\rangle$8 are provably hard to distinguish using any poly-copy protocol (Haug et al., 25 Apr 2025).

The interplay of magic and entropy underpins cryptographic security: so-called "pseudomagic" ensembles can only hide extensive magic when the entropy is also extensive—entropy is a necessary resource to conceal magic from any adversary. Thus the capacity to mimic high-magic states using low-magic states requires an extensive entropy budget, establishing entropy as a cryptographic resource analogous to magic (Haug et al., 25 Apr 2025).

6. Connections to Fault Tolerance and Quantum Resource Theories

Resource theories of magic treat the stabilizer convex hull and stabilizer-preserving operations as free. Robustness measures inherit properties such as faithfulness (zero iff stabilizer), convexity, monotonicity under free operations, and often submultiplicativity (Howard et al., 2016, Warmuz et al., 2024). Channels and many-body extensions can be analyzed by their channel robustness and capacity, providing compositional bounds for simulation and synthesis (Seddon et al., 2019, Saxena et al., 2022).

Geometric approaches, such as the Minkowski-function magic monotone, provide computationally efficient, necessary, and sufficient criteria using integer hyperplane optimizations (Warmuz et al., 2024).

7. Summary Table: Core Quantitative Findings

Regime / Observable	Robustness Threshold / Scaling Law	Paper
Global depolarizing noise ($|S_i\rangle$9 qubits)	Magic persists if $R(\rho)=1$0	(Haug et al., 25 Apr 2025)
Local depolarizing, random circuits	Critical depth $R(\rho)=1$1; $R(\rho)=1$2 independent of $R(\rho)=1$3	(Haug et al., 25 Apr 2025)
C$R(\rho)=1$4Z hypergraph states (noise)	Magic threshold scales as $R(\rho)=1$5	(Wei et al., 2024)
3-complete hypergraph states (global)	Nonvanishing threshold $R(\rho)=1$6 as $R(\rho)=1$7	(Wei et al., 2024)
Ising chain, mutual RoM at $R(\rho)=1$8	No sudden death; robust for all $R(\rho)=1$9	(Timsina et al., 17 Jul 2025)
Random circuits (IonQ experiments)	Magic detectable for single T-gate even at $\rho$0	(Haug et al., 25 Apr 2025)

• (Howard et al., 2016) Application of a resource theory for magic states to fault-tolerant quantum computing
• (Seddon et al., 2019) Quantifying magic for multi-qubit operations
• (Heinrich et al., 2018) Robustness of Magic and Symmetries of the Stabiliser Polytope
• (Hamaguchi et al., 2023) Handbook for Quantifying Robustness of Magic
• (Wei et al., 2024) Noise robustness and threshold of many-body quantum magic
• (Warmuz et al., 2024) A magic monotone for faithful detection of non-stabilizerness in mixed states
• (Haug et al., 25 Apr 2025) Efficient witnessing and testing of magic in mixed quantum states
• (Timsina et al., 17 Jul 2025) Robustness of Magic in the quantum Ising chain via Quantum Monte Carlo tomography
• (Saxena et al., 2022) Quantifying dynamical magic with completely stabilizer preserving operations as free
• (Niroula et al., 2023) Phase transition in magic with random quantum circuits

For a comprehensive treatment of the operational, computational, and physical aspects of magic robustness—including rigorous thresholds, experimental validation, property testing, and resource-theoretic structures—see (Haug et al., 25 Apr 2025, Hamaguchi et al., 2023), and (Wei et al., 2024).